Topological Analysis

    It has been shown that topological analysis of chemical structures can reveal a correlation between a structure and its phys-chemical properties. Usually topological analysis is quite laborious due to the amount of paid concentration and people trying to avoid it. However, the conclusions, which can be made from the analysis, can reveal interesting structural particularities of the structure and hence the analysis is an essential part of the structural description.
    Normally, topological analysis consists of two parts – construction of topological network and the analysis itself. The analysis is the description of a structure in terms of connectivity, which can be expressed either in terms of vertex symbols or other characteristics of a network or in a descriptive way. Construction of topological networks becomes controversial when complex systems are considered. Even if structural connectivity is well-defined, the selection of “real” topological nodes, which carry chemical sense, is not always straightforward. A node can be defined as a single unit connecting at least three others (of the same or different kind). If we consider extended coordination networks, then these nets usually are modular, e.g. consist of distinct ligand and metal centres and the selection of nodes can be done easily. However, if pure inorganic networks are considered, such as interlinked clusters, then the node selection procedure becomes much more complicated. Original algorithm, we used for the selection of topological nodes, was designed mainly for coordination networks and was based on a single atom or atom type. However, the approach becomes useless, when analysed network contains clusters or several components. Recent modification of the algorithm does not need a selection of nodes and based purely on the definition of a topological node and a very simple mechanism is provided to give chemical sense to a network.
    I have provided two short clips showing the procedure of the analysis. I tried to develop an algorithm for automatic identification of nodes in chemical sense by analysing quite complicated networks consisting of clusters. But so far I can give only a few tips:

After the construction of a topological network:
1. metal-ligand and pure organic frameworks
    a. identify all topological bonds (nodes) belonging only to the a single ligand (coordination bonds, and bonds coming up from the “edge-effect” should be omitted) and execute the “collide to one node” command.
2. clusters
    a. identify the content of a cluster and its main component (for example Mo for the examples below). Check if the main component does non bind clusters directly. If it does not connect clusters, then “collide” all the bonds belonging to the main component (otherwise choose different component). Now check if the nodes, formed by the “collision” can be collided further. If not then try to identify another network component which does not connect clusters directly and perform the “collision”.

The complete list of topological bonds can be accessed from the main toolbar. Also you can use on-screen selections to get a list of topological bonds. Just be sure that only topological bonds are in the list: in this case the commands for topological bonds become accessible.

Note that each “collision” step must simplify the network. However, the 3D structure of the network must be also maintained. If one of the steps degenerates the network, start from the construction of the network (undo does not work). Also, if necessary, use different groups of bonds to identify main components.

The "edge-effect" appears when asymmetric unit of a structure contains parts of ligands. It appears as strange bond located mainly on the "surface" of the packing. The network construction algorithm partially removes the edge-effect, and works alright in tested cases. However if you have noticed its malfunctioning, please let me know.

Click here for examples of analysis of clusters.

AVI Clips: (800x600, zip)
Analysis of a coordination network [1].
Analysis of a discrete structure [2].

1. O. V. Dolomanov, D. B. Cordes, N. R. Champness, A. J. Blake, L. R. Hanton, G. B. Jameson, M. Schröder, and C. Wilson, Chem. Commun., 2004, p. 642-643.
2. O. V. Dolomanov, A. J. Blake, N. R. Champness, C. Wilson, and M. Schröder, Chem. Commun., 2003, p. 682-683.